基于颜色差异计算的图像显著性检测算法研究

针对目前显著性算法普遍面临的背景噪声较多,准确性较低等问题,提出一种基于图像颜色对比的显著性检测算法.首先在Lab伪彩色空间域将图像进行SLIC超像素分割,依照图像的全局对比计算,确定前景预选区域.然后以该区域内超像素特征通道信息值作为参考值,计算各超像素显著值,得到初步显著图.最后构建视觉中心,考虑颜色权重信息对显著...

Research on Image Saliency Detection Algorithm Based on Color Difference Calculation

Aiming at the problems that the current saliency algorithms generally face are more background noise and lower accuracy,a saliency detection algorithm based on image color contrast is proposed. First,the image is divided into SLIC superpixels in the Lab pseudo-color space domain,and the foreground preselected area is determined according to the global comparison calculation of the image. Then,using the superpixel feature channel information value in the area as a reference value,the superpixel saliency value is calculated to obtain a preliminary saliency map. Finally,a visual center is built,the influence of color weight information on the significant target area is considered,and it is optimized to obtain the final result. The current general data set is used for testing and compared with the other 6 mainstream algorithms. The results show that the experimental algorithm in this paper not only has better final result graph quality,but also performs well on multiple objective evaluation indicators.